Functional analyses of small secreted cysteine‐rich proteins identified candidate effectors in Verticillium dahliae

Abstract Secreted small cysteine‐rich proteins (SCPs) play a critical role in modulating host immunity in plant–pathogen interactions. Bioinformatic analyses showed that the fungal pathogen Verticillium dahliae encodes more than 100 VdSCPs, but their roles in host–pathogen interactions have not been fully characterized. Transient expression of 123 VdSCP‐encoding genes in Nicotiana benthamiana identified three candidate genes involved in host–pathogen interactions. The expression of these three proteins, VdSCP27, VdSCP113, and VdSCP126, in N. benthamiana resulted in cell death accompanied by a reactive oxygen species burst, callose deposition, and induction of defence genes. The three VdSCPs mainly localized to the periphery of the cell. BAK1 and SOBIR1 (associated with receptor‐like protein) were required for the immunity triggered by these three VdSCPs in N. benthamiana. Site‐directed mutagenesis showed that cysteine residues that form disulphide bonds are essential for the functioning of VdSCP126, but not VdSCP27 and VdSCP113. VdSCP27, VdSCP113, and VdSCP126 individually are not essential for V. dahliae infection of N. benthamiana and Gossypium hirsutum, although there was a significant reduction of virulence on N. benthamiana and G. hirsutum when inoculated with the VdSCP27/VdSCP126 double deletion strain. These results illustrate that the SCPs play a critical role in the V. dahliae–plant interaction via an intrinsic virulence function and suppress immunity following infection.


| INTRODUC TI ON
Some fungi have evolved the ability to cause plant diseases and are a threat to many economically important crops. Although plants lack mobile defence cells or adaptive immune systems, they have evolved an innate immune system to recognize and respond to pathogens (Cook et al., 2015). Innate immunity in plants involves the induction of cascading antimicrobial responses following recognition of a pathogen, including rapid changes in calcium ion levels, a reactive oxygen species (ROS) burst, callose deposition, cell wall reinforcement, and increases in defence-related gene expression (Boller and Felix, 2009;Zipfel, 2009). Plants can detect pathogens via different immunogenic signals: pathogen-or microbe-associated molecular patterns (PAMP/ MAMP) (Jones and Dangl, 2006;Zipfel, 2008). Recognition of PAMPs by plant cell surface-localized pattern recognition receptors (PRRs) triggers defence responses known as PAMP-triggered immunity (PTI). To surmount plant innate immunity, pathogens may deliver effector molecules either to the host apoplast or cytoplasm, resulting in effector-triggered susceptibility (ETS).
Pathogen effector proteins may also be recognized by plant R proteins, which activate a second layer of immunity known as effector-triggered immunity (ETI) (Jones and Dangl, 2006;Zipfel, 2008;Dodds and Rathjen, 2010;Giraldo and Valent, 2013). From the point of view of a pathogen, secreted effectors play a critical role in successful infection through multiple pathways and especially by suppressing plant defence responses (Toruño et al., 2016).
Since the molecular cloning of the first fungal effector from Cladosporium fulvum (van Kan et al., 1991), numerous effectors have been characterized and investigated in many plant pathogens (Houterman et al., 2008;Stergiopoulos and de Wit, 2009). Features that discriminate fungal effectors from secreted non-effectors are predominantly sequence length, molecular weight, net charge, as well as cysteine, serine, and tryptophan content (Sperschneider et al., 2016). The effectors generally present low sequence similarity to known proteins, domains, and motifs (Rep, 2005). Therefore, certain features are predictive of potential candidate effectors in fungi, including the presence of signal peptides, protein size, the lack of transmembrane domains or glycosyl-phosphatidylinositol (GPI)-anchor sites, cysteine-richness, and localization in the host plant (Rep, 2005;Stergiopoulos and de Wit, 2009;Kim et al., 2016). In particular, small cysteine-rich proteins (SCPs) are typical of the known apoplastic effectors (Rep, 2005;Stergiopoulos and de Wit, 2009;Lu and Edwards, 2016;Qi et al., 2016;Cheng et al., 2017) in which cysteine residues form disulphide bonds that appear to enhance effector stability in the apoplastic space, which is rich in proteases (Kamoun, 2006;Stergiopoulos and de Wit, 2009;Saunders et al., 2012). For instance, Rhynchosporium secalis Nip1 encodes a small protein with 10 cysteine residues and five intramolecular disulphide bridges (Rohe et al., 1995). Small proteins that are rich in cysteine residues and predicted to be secreted have been proposed as likely candidates for effector proteins that play a critical role during host-pathogen interactions.
Verticillium dahliae is a notorious fungal pathogen that attacks a wide range of hosts, targets the xylem tissue, and causes verticillium wilt disease on many economically important crops (Klosterman et al., 2009;Inderbitzin and Subbarao, 2014). Previous studies have shown that secreted proteins probably play a critical role in the pathogenesis of V. dahliae (Fradin and Thomma, 2006;Chu et al., 2015;Chen et al., 2016a;2016b;Zhang et al., 2018), and the function or the activity of several secreted proteins have been elucidated (de Sain and Rep, 2015;Klimes et al., 2015;Wang et al., 2018;Zhang et al., 2018). Similar to the mechanisms discovered in other phytopathogens, V. dahliae secretes proteins to manipulate host immunity during infection (de Jonge et al., 2012;Zhou et al., 2012;Santhanam et al., 2013;Liu et al., 2014;Gui et al., 2017Gui et al., , 2018Kombrink et al., 2017;Zhang et al., 2017;Qin et al., 2018). Many of these effectors are relatively small and enriched in cysteine residues, including Ave1, VdCBM1, and VdSCP7 (VDAG_07157) (de Jonge et al., 2012;Gui et al., 2017;Zhang et al., 2017). For instance, the avirulence gene Ave1 encodes an SCP preprotein of 134 amino acids and contains four cysteine residues in the mature protein (de Jonge et al., 2012). Complete genome sequencing has greatly facilitated the rapid identification of genes encoding SCPs. Data mining of the V. dahliae secretome showed that the VdLs.17 genome encodes more than 100 hypothetical proteins that were designated as small (≤400 amino acids), cysteine-rich (≥4 cysteine residues) proteins, and some of these potentially function as effectors (Klosterman et al., 2011). Our recent study of the V. dahliae Vd991 (from Gossypium hirsutum) genome identified 127 SCPs (Chen et al., 2018) but the functions of SCPs in V. dahliae have remained largely unknown.
In the current study, all 127 SCPs encoded in the genome of V. dahliae Vd991 were investigated for their activity in plantpathogen interactions using a transient expression system in Nicotiana benthamiana. N. benthamiana was the system of choice for this work because it is fast growing, highly susceptible to V. dahliae (Wright and Biss, 1968), and a model host species to screen for gene functions through leaf agroinfiltration (Bally et al., 2018).
The main objectives of this study were to (a) characterize the features of SCPs in V. dahliae Vd991; (b) determine specific SCPs that trigger an immune response in N. benthamiana using a transient expression assay; (c) study the potential role of disulphide bridges in SCP-triggered immunity; (d) test whether other SCPs could suppress immunity; and (e) investigate the virulence function of three SCPs in cotton and N. benthamiana that could induce the immune responses.

| Identification of genes encoding VdSCPs in the V. dahliae genome
In total 739 genes were predicted to encode secreted proteins in the genome of V. dahliae (strain Vd991). Among these, 127 hypothetical proteins encoding SCPs (<400 amino acids, ≥4 cysteine residues here described as VdSCPs) were predicted as effectors (Chen et al., 2018). These VdSCPs contain a signal peptide but lack transmembrane domains (Table S1), typical characteristics of proteins transported to the extracellular space, where they may be involved in host-pathogen interactions. Four VdSCPs containing four cysteine residues (VdSCR3, VdSCR46, VdSCR57, and VdSCR59) were excluded from the set of predicted SCPs due to the presence of one cysteine residue within the signal peptide, which would result in mature proteins with only three cysteine residues (Tables S2 and S3). Physical location analysis showed that the VdSCPs are distributed across all eight chromosomes in Vd991 (Figure 1a and Table S3), and their delineation on the chromosomes from Vd991 was determined by chromosomal synteny with the reference genome of strain JR2 (Chen et al., 2018). Chromosome 1 harbours the largest number (29 in total) of VdSCPs ( Figure 1a and Table S3). Most VdSCPs are between 100 and 300 amino acids and contained 2%-4% cysteine residues (Figure 1b and Table S3). To further understand the characteristics of VdSCPs, the distribution pattern of cysteine residues for each protein was investigated by the interval length between the neighbouring cysteine residues, revealing that the cysteine residues in some VdSCPs have a similar distribution pattern ( Figure 1c and Table S4). For instance, nine VdSCPs have similar distribution patterns of cysteine residues C-X 3 -C-X 9-12 -C-X 6-7 -C-X 1 -C-X 11-14 -C-X 4-6 -C-X 14-16 -C (Figure 1c and   Table S4), and are composed of the common domain (IPR008427, CFEM domain) in several fungal extracellular membrane proteins (Table S5). However, the distribution of cysteine residues in most VdSCPs (101 in total) did not follow this pattern. In addition, comparative analysis of VdSCPs to the SCPs encoded in the V. dahliae genomes from strains JR2 and VdLs.17 showed that 10 VdSCPs were specific to Vd991 (Table S3) (Figures 2a and S1). Although immunoblotting analysis confirmed the effective translation of four additional arbitrarily selected VdSCPs (VdSCP2, VdSCP47, VdSCP74, and VdSCP90), these failed to trigger cell death in N. benthamiana (Figure 2a,b). These results indicate that VdSCP27, VdSCP113, and VdSCP126 in particular possess cell death-inducing activities in N. benthamiana. VdSCP27,VdSCP113,and VdSCP126 encode proteins of 214,191, and 212 amino acids in length, and contain four, seven, and four cysteine residues, respectively (Table S3). BlastP analysis revealed proteins with no known functions homologous to these three VdSCPs that encode hypothetical proteins, and only VdSCP27 contains a conserved domain (IPR025649, protein of unknown function DUF4360) as predicted by InterProScan (Table S5).

| VdSCP27, VdSCP113, and VdSCP126 mainly localize on the periphery of N. benthamiana cells
To test the secretory function of VdSCP27, VdSCP113, and VdSCP126, the yeast signal trap assay system based on the requirement of yeast cells for invertase secretion to grow on sucrose or raffinose media was used for the functional analysis of the putative N-terminal signal peptide (Liu et al., 2013; Table S1). Like the known functional signal peptide Avr1b, the signal peptide of all three VdSCPs had the ability to mediate the secretion of invertase after each signal peptide region was fused with the invertase (Gu [2, 4) means the cysteine percentage is greater than or equal to 2% and less than 4%. (c) Calculation of protein numbers according to the number and percentage of cysteine residues, respectively. (d) Analysis of the conserved distribution pattern of the cysteine residues of the VdSCPs. X represents the random amino acid residue, and the subscript numbers represent the total number of amino acid residues between the two cysteines 2.4 | VdSCP27, VdSCP113, and VdSCP126 trigger immunity in N. benthamiana by inducing a diversity of defence responses To examine whether VdSCP27, VdSCP113, and VdSCP126 act as effectors to induce defence responses during host-pathogen interactions, several types of defence responses were assayed in N. benthamiana leaves following agroinfiltration, including the ROS accumulation, callose deposition, activation of defence response genes, and restriction of disease development. ROS accumulation was substantially enhanced 3 days after agroinfiltration of the constructs encoding VdSCP27, VdSCP113 or VdSCP126, and the known PAMP VdEG3 (Gui et al., 2017) compared to the control agroinfiltrated with the GFP constructs (Figure 4a,c). N. benthamiana leaves F I G U R E 2 Identification of cell deathinducing activity of Verticillium dahliae secreted small cysteine-rich proteins (VdSCPs) in Nicotiana benthamiana. (a) Induction of cell death was detected for VdSCPs in N. benthamiana leaves from 4-week-old plants at 6 days after infiltration with Agrobacterium tumefaciens expressing the indicated genes. The construct expressing green flourescent protein (GFP) was used as a negative control. (b) Immunoblotting analysis of four non-cell death-inducing activity of VdSCPs (VdSCP1, VdSCP47, VdSCP74, and VdSCP90) fused to the FLAG-tag in N. benthamiana leaves 60 hr after infiltration. Ponceau S-stained RuBisCO protein is a total protein loading control displayed strong callose deposition after agroinfiltration with constructs encoding VdSCP27, VdSCP113 or VdSCP126 and the known PAMP VdEG3, compared to the control GFP (Figure 4b,d).
Furthermore, several genes associated with immunity were significantly induced after agroinfiltration with the constructs encoding VdSCP27, VdSCP113 or VdSCP126, including hypersensitive F I G U R E 3 Subcellular localization of the Verticillium dahliae proteins VdSCP27, VdSCP113, and VdSCP126. (a) Functional validation of the putative N-terminal signal peptide of VdSCP27, VdSCP113, and VdSCP126 using the yeast invertase secretion assay (Liu et al., 2013). The yeast strain YTK12 could not grow on CMD-W medium (i.e., without tryptophan). The strain containing the pSUC2 vector grew based on the function of the Trp operon. Fusion of the functional signal peptide of each of the three VdSCPs in-frame with mature yeast invertase enabled secretion of invertase, resulting in growth on YPRAA medium. The known functional signal peptide Avr1b was used as a positive control. (b) Subcellular localization of VdSCP27, VdSCP113, and VdSCP126 was determined by transient expression of C-terminal green fluorescent protein (GFP)tagged proteins in Nicotiana benthamiana leaves. The fluorochrome FM4-64 was used for dyeing the cell membrane. Three genes were transiently expressed in 4-week-old N. benthamiana leaves and harvested at 2 days after agroinfiltration. The vector pBin-GFP was used as negative control. Fluorescence was scanned using a Leica TCS SP8 confocal microscopy system using 200× magnification with an excitation wavelength at 488 nm and emission at 510 nm for GFP, an excitation wavelength at 543 nm and emission of 562 nm for FM4-64. Bars = 35 μm. (c) Functional signal peptides are required for cell death-inducing activities of VdSCP27, VdSCP113, and VdSCP126 in N. benthamiana. Deletion of the signal peptide encoding region in VdSCP27, VdSCP113, and VdSCP126 (yielding the products VdSCP27 ∆SP , VdSCP113 ∆SP , and VdSCP126 ∆SP ) resulted in no induction of cell death in 4-week-old plants at 6 days after infiltration. Positive and negative control constructs expressed Bcl-2-associated X protein (BAX) and GFP, respectively. (d) Immunoblotting analysis of transiently expressed VdSCP27 ∆SP , VdSCP113 ∆SP , and VdSCP126 ∆SP fused to the FLAG-tag in N. benthamiana leaves at 60 hr after infiltration. The expected sizes of VdSCP27 ∆SP , VdSCP113 ∆SP , and VdSCP126 ∆SP were 21.2, 17.9, and 20.4 kDa, respectively. Ponceau S-stained RuBisCO protein is a total protein loading control response genes (HSR203 and HIN1), the PTI-associated marker gene GRAS2, and defence response genes associated with the salicylic acid signalling pathway (PR1 and GLNb) and the jasmonic acid/ ethylene signalling pathway (PR4) (Figure 4e). In contrast, GFP could not induce the expression of resistance genes compared with the positive control VdEG3 (Figure 4c). In addition, agroinfiltration of the VdSCP27, VdSCP113 or VdSCP126 constructs into N. benthamiana leaves resulted in a significant reduction of lesion area 48 hr after inoculation with Botrytis cinerea, compared to agroinfiltration of GFP alone (Figure 4f,g). Together, these results suggested that VdSCP27, VdSCP113, and VdSCP126 can each act as elicitors that induce cell death by triggering defence responses during host-pathogen interaction in N. benthamiana.

| Cysteine residues are essential for the function of VdSCP126, but not VdSCP27 and VdSCP113
VdSCP27, VdSCP113, and VdSCP126 are enriched in cysteine residues that may be involved in disulphide bond formation. Bioinformatic analysis predicted two, three, and two cysteine disulphide bonds in VdSCP27, VdSCP113, and VdSCP126 ( Figure 6), respectively, although the scores were relatively low for confirmation of some of these disulphide bonds. To test whether the cysteine residues in VdSCP27, VdSCP113, and VdSCP126 were essential to trigger immunity, mutant alleles of disulphide bonds in these VdSCPs were generated in which all cysteine residues were replaced with alanine (C-A). Transient expression of these mutant alleles demonstrated that substitution of all cysteine residues in VdSCP27 and VdSCP113 retained cell deathinducing activity in N. benthamiana but led to a complete loss of cell death-inducing activity in the VdSCP126 mutant allele (Figure 6a,c,e).
In addition, when single cysteine residues (C140A, C172A, C201A or C211A) were substituted in VdSCP126, the corresponding mutant allele did not induce cell death for up to 5 days after agroinfiltration in N. benthamiana leaves (Figure 6e). Immunoblotting analysis confirmed that all the mutant alleles of these three VdSCPs were successfully expressed 3 days after infiltration in N. benthamiana leaves (Figure 6b,d,f).
These results suggested that cysteine residues are critical for inducing cell death in VdSCP126, but not in VdSCP27 and VdSCP113. We F I G U R E 4 Identification of the immunity triggered by VdSCP27, VdSCP113, and VdSCP126 in Nicotiana benthamiana. (a) Reactive oxygen species (ROS) accumulation after transient expression of VdSCP27, VdSCP113, and VdSCP126 in N. benthamiana leaves. ROS accumulation was assessed in N. benthamiana leaves from 4-week-old plants expressing VdSCP27, VdSCP113, and VdSCP126 by staining with 3,3′-diaminobenzidine. Expression of the known pathogen-associated molecular pattern (PAMP) VdEG3 and green fluorescent protein (GFP) were included as positive and negative controls, respectively. (b) Detection of callose deposition by aniline blue straining. N. benthamiana leaves from 4-week-old plants were analysed 2 days after agroinfiltration with expression constructs encoding VdSCP27, VdSCP113, and VdSCP126. VdEG3 and GFP were used for positive and negative control constructs, respectively. (c) Detection of ROS induction levels associated with expression of VdSCP27, VdSCP113, and VdSCP126 after infiltration in N. benthamiana. Values are means ± SD from three independent experiments. Asterisks indicate a significant difference (p < .01) relative to the control of agroinfiltrated GFP expression construct in an unpaired Student's t test. (d) Average number of callose deposits/mm 2 on expressing VdSCP27, VdSCP113, and VdSCP126. Values represent the means ± SE of three independent samples. Asterisks indicate a significant difference (p < .01) relative to the control of agroinfiltration GFP according to an unpaired Student's t test. (e) The induction defence response genes triggered by expression of VdSCP27, VdSCP113, and VdSCP126. The transcripts were detected by quantitative reverse transcription PCR in 4-week-old N. benthamiana leaves 24 hr after agroinfiltration of constructs encoding VdSCP27, VdSCP113, and VdSCP126, and GPF as control.
). (f) Defence response induced by recombinant VdSCP27, VdSCP113, and VdSCP126 proteins. N. benthamiana leaves from a 5-week-old plant were pretreated with 100 mM of the indicated recombinant protein and inoculated 12 hr later with 5 µl of 2 × 10 6 B. cinerea conidia/ml. Lesions were observed at 2 days post-inoculation (dpi). (g) Lesion development of B. cinerea on N. benthamiana leaves was evaluated beginning 2 dpi by determining the average lesion diameter on six leaves from six plants each. Error bars represent standard errors of the mean. Values represent the averages of three independent measurements with three replicates each. Error bars represent standard errors of the mean. Asterisks ( * ) and double asterisks ( * * ) represent statistical significance at p < .05 and p < .01, respectively, based on unpaired Student's t tests therefore speculate that the disulphide bonds probably are essential for the biological function of VdSCP126 during host-pathogen interactions, but the position of disulphide bonds that confers this function requires further scrutiny.

| V. dahliae employs VdSCP effectors to suppress immunity triggered by VdSCP27, VdSCP113, and VdSCP126
To surmount immunity triggered by VdSCP27, VdSCP113, and VdSCP126, V. dahliae probably delivers other SCP effectors to promote plant susceptibility by suppressing plant immune responses. To test this hypothesis in V. dahliae, we selected a set of potential VdSCP effectors (VdSCP23, VdSCP38, VdSCP58, VdSCP60, VdSCP65, and VdSCP70) that suppressed cell death induced by the PAMP VdEG1 (unpublished data), and which trigger cell death and PTI in N. benthamiana (Gui et al., 2017). All six VdSCP effectors that could suppress VdEG1-triggered cell death also suppressed cell death triggered by VdSCP27 and VdSCP113 (Figure 7a), but could not completely suppress cell death triggered by VdSC126 (Figure 7a).
VdCBM1 is a known effector that can suppress PTI during infection in V. dahliae (Gui et al., 2017). We further tested whether VdCBM1 had the ability to suppress the immunity triggered by VdSCP27, VdSCP113, and VdSCP126. VdSCP27, VdSCP113, and VdSCP126 were each separately expressed with VdCBM1 in N. benthamiana by co-agroinfiltration. Expression of VdCBM1 suppressed the immune response when compared to the normal cell death-inducing activities associated with constructs encoding VdSCP27, VdSCP113, and VdSCP126 ( Figure 7a). The VdCBM1-encoding protein also contains four cysteine residues and was characterized previously as a SCP (Klosterman et al., 2011;Gui et al., 2017). These results suggested that cell death triggered by VdSCP27, VdSCP113, and VdSCP126 is differentially and effectively suppressed by VdSCP effectors in N. benthamiana. In addition, conductivity assays revealed that coexpression of VdSCP27, VdSCP113, and VdSCP126 separately with the GFP expression cassette induced electrolyte leakage 3 days after co-agroinfiltration in N. benthamiana leaves, but the electrolyte leakage was significantly reduced in the areas of co-agroinfiltration with VdSCP27, VdSCP113, and VdSCP126 constructs in combination with the VdCBM1 construct ( Figure 7b). Correspondingly, the defence genes induced by the expression of VdSCP27, VdSCP113, or VdSCP126 were significantly suppressed when co-expressed with VdCBM1 in N. benthamiana leaves (Figure 7d,e,f). In addition, ROS accumulation in N. benthamiana leaves was significantly impacted with co-expression of VdCBM1 (Figure 7c). To further detect whether VdSCP27, VdSCP113 or VdSCP126 were functional or exhibit quantitative redundancy during F I G U R E 6 Prediction and functional analysis of disulphide bonds in the Verticillium dahliae proteins VdSCP27, VdSCP113, and VdSCP126. Disulphide bond predictions for VdSCP27 (a), VdSCP113 (c), and VdSCP126 (e) were conducted using the web-based program DiANNA (http://clavi us.bc.edu/~clote lab/DiANN A/). Values represent the score of cysteine residues likelihood of involvement in disulphide bond formation. SP, signal peptide. (a), (c), (e) The function of cysteine residues and the importance of disulphide bond formation were detected by cell death induction assays in 4-week-old Nicotiana benthamiana leaves after agroinfiltration of wild-type genes and cysteine residue mutant alleles. The leaf phenotypes were photographed 6 days after agroinfiltration. "C-A" indicates all cysteine residues were replaced by alanine; "C-No.-A" indicates the single cysteine residues were replaced by alanine in the respective position in VdSCPs. The efficiency of transient expression of the wild-type genes and cysteine residue mutant alleles encoding the V. dahliae proteins VdSCP27 (b), VdSCP113 (d), and VdSCP126 (f) were validated by immunoblotting analysis 60 hr after agroinfiltration. in N. benthamiana. Ponceau S-stained RuBisCO protein is shown as a total protein loading control infection of host plants, double deletion mutants of two VdSCP pairs (∆SCP27_113, ∆SCP27_126, ∆SCP113_126) were constructed ( Figure   S5). The pathogenicity assays showed that deletion of VdSCP27 and VdSCP126 together impaired virulence toward both N. benthamiana and cotton plants (Figure 8a,b). Quantification of fungal biomass in inoculated N. benthamiana and cotton also revealed obvious fungal biomass reduction in deletion strain ∆SCP27_126 compared with the wild-type strains (Figure 8c,d). However, the other double deletion mutants (∆SCP27_113 and ∆SCP113_126) did not impair the virulence towards the two hosts. These results indicate that VdSCP27 and VdSCP126 also act together as virulence factors in N. benthamiana and cotton interactions with V. dahliae. (a) Assay for suppression of cell death-inducing activity of six VdSCP effectors and VdCBM1 by co-agroinfiltration with expression constructs encoding VdSCP27, VdSCP113, and VdSCP126 in 4-week-old Nicotiana benthamiana leaves. Co-agroinfiltration of constructs encoding six VdSCP effectors. VdCBM1 and green fluorescent protein (GFP) were expressed for controls. Photographs were taken 6 days after agroinfiltration. (b) Electrolyte leakage assay in 4-week-old N. benthamiana leaves 48 hr after co-infiltration of constructs expressing VdSCP27, VdSCP113, and VdSCP126 simultaneously with VdCBM1. (c) 3,3′-diaminobenzidine (DAB) staining of the inhibition of reactive oxygen species accumulation 6 days after co-agroinfiltration of the VdCBM1 construct with those of VdSCP27, VdSCP113, and VdSCP126. Co-agroinfiltration of VdSCPs alone and GFP were used as controls. Agroinfiltration of empty vector (pGR107) served as a control (Mock). Quantitative reverse transcription PCR analyses of the inhibition of defence response gene expression 3 days after co-agroinfiltration of expression constructs VdCBM1 with VdSCP27 (d), VdSCP113 (e), and VdSCP126 (f). Co-agroinfiltration of VdSCPs alone and GFP were used as controls. Blue columns represent expression of defence-related genes in response to each of the VdSCPs. Red columns represent the expression of defence-related genes when VdSCPs were co-infiltrated with VdCMB1. Values represent the averages of three independent measurements with three replicates each, and error bars represent standard errors of the mean. Asterisks ( * ) and double asterisks ( * * ) represent statistical significance at p < .05 and p < .01, respectively, according to unpaired Student's t tests strain Vd991 from cotton was undertaken. This involved the screening of 123 putative small cysteine-rich secreted proteins (Chen et al., 2018)  Comparative genomics analyses have suggested that V. dahliae secretes over 700 proteins, including more than 100 SCPs with unknown function (hypothetical protein) that are thought to function in the host apoplast to promote disease (Klosterman et al., 2011;Chen et al., 2018). In our previous study, we found that the genome of V. dahliae strain Vd991 encodes 127 SCPs of unknown function (Chen et al., 2018), and confirmed that 123 encode SCPs based on their mature proteins (Table S3). With the large-scale transient expression of these 123 VdSCPs, we first identified three new SCPs, VdSCP27, VdSCP113, and VdSCP126, that induced cell death in N. benthamiana (Figures 1 and S1). VdSCP27, VdSCP113, and VdSCP126 localized along the N. benthamiana cell periphery, possibly acting as PAMPs to induce cell death depending on their signal transduction by the BAK1/SOBIR1 complex (Figures 3 and   5). However, the subcellular location also showed that the three VdSCPs, especially VdSCP27 and VdSCP113, can also be located in cytoplasm, such as on the cytosolic bridges (Figure 3), suggesting that these three effectors also may have an intracellular role or are internalized similar to the effector SsSSVP1 (Lyu et al., 2016).

| D ISCUSS I ON
Only the A. tumefaciens pBin-GFP was used as a negative control along with a cell membrane colocalizing dye. Additional controls that accumulate in the same cell compartments with target proteins (VdSCPs) could enhance our understanding of the accuracy of the subcellular locations of the three VdSCPs in future studies.
One obvious and common feature of SCPs, as well as some effectors that maintain structural stability in oxidizing environments, is the presence of multiple cysteine residues and their disulphide bonds (Sevier and Kaiser, 2002). The intramolecular disulphide bond is presumed to be important for protein folding, stability, and protection of such proteins in the harsh acidic and protease-rich apoplast when they are delivered during plant infection (Kamoun, 2006). Disulphide bonds that are important for protein function have been identified in several effectors, including Avr2, Avr4, and Avr9 of Cladosporium fulvum (van den Hooven et al., 2001;Luderer et al., 2002;van den Burg et al., 2003;Van't Klooster et al., 2011), SsSSVP1 from Sclerotinia sclerotiorum (Lyu et al., 2016), SnTox1 produced by Stagonospora nodorum (Liu et al., 2012), and Pep1 from Ustilago maydis (Doehlemann et al., 2009). In some cases, the post-translational modification is also believed to be important for biological activity of SCPs, as in the modification of cysteine residues by S-nitrosylation (Ling et al., 2017). In the current study, mutational analysis showed that all cysteine residues are required for VdSCP126 to trigger cell death and immunity in N. benthamiana ( Figure 6), suggesting that disulphide bonds formed by all cysteine residues probably are also important for the host infection mediated by VdSCP126. Unexpectedly, substitution of all cysteine residues in VdSCP27 and VdSCP113, however, did not affect the ability of each to induce cell death in N. benthamiana (Figure 6). The immunoblotting of SCP47 and SCP90 revealed that the actual sizes did not match the theoretical sizes, and these changes probably were caused by post-translational modifications (Figure 2b). A few studies have characterized the lack of involvement of SCP cysteine residues in host-pathogen interactions. The role of cysteines residues in VdSCP27 and VdSCP113 will require further clarification.
During the coevolution of host-microbe interactions, pathogens acquired the ability to deliver effector proteins to interfere with PTI, enabling pathogens to infect their respective host plants and cause disease (Chisholm et al., 2006;Jones and Dangl, 2006). For instance, Puccinia striiformis secretes a candidate effector, PSTha5a23, which mediates plant defence suppression and promotes wheat rust (Cheng et al., 2017). Similarly, the immunity triggered by VdSCP27, VdSCP113, and VdSCP126 in a BAK/ SOBIR1-dependent manner can also be suppressed by several VdSCP effectors (Figure 6a). In our previous study, we found that V. dahliae can deliver the effector VdCBM1, a protein that also contains four cysteine residues and is a secreted SCP that suppresses PTI (Gui et al., 2017). VdCBM1 can also suppress the immunity triggered by damage-associated molecular patterns (DAMPs), including the products from VdCUT11 . In this study, we rediscovered that the immunity triggered by VdSCPs in a BAK1/ SOBIR1-dependent manner is also suppressed by VdCBM1 ( Figures   5 and 7). These results indicate that the V. dahliae VdSCPs could be part of a network that participates in host-pathogen interaction to surmount the immunity triggered by VdSCPs and promote colonization during infection. Some SCPs are required for full virulence in many pathogens that have been analysed (Bolton et al., 2008;Doehlemann et al., 2009;Chen et al., 2016b;Qin et al., 2018). In our study, we found that deletion of VdSCP27, VdSCP113, or VdSCP126 independently did not affect the virulence of V. dahliae in tobacco and cotton plants ( Figure   S4). These results conflicted with our initial expectation that the immunity triggered by these VdSCPs is eliminated in the corresponding gene deletion strains, resulting in enhanced virulence during host infection, as deletion of immunity factor VdSCP7 (VDAG_07157) significantly enhanced virulence on cotton . Interestingly, analyses of the double VdSCP deletion mutant showed that VdSCP27 and VdSCP126 together act as virulence factors in V. dahliae (Figure 8).
In Phytophthora sojae, glycoside hydrolase 12 protein is a major virulence factor during soybean infection and is recognized as a PAMP (Ma et al., 2015), and so are VdEG1 and VdEG3 from V. dahliae (Gui et al., 2017), suggesting that VdSCP27 and VdSCP126 function in virulence, although they also may be recognized by the host plants to induce immunity. Some fungal effectors are often considered dispensable for virulence due to functional or quantitative redundancy (Sharpee and Dean, 2016), as many PAMPs have been found in V. dahliae (de Jonge et al., 2012;Zhou et al., 2012;Santhanam et al., 2013;Gui et al., 2017; and deletion of VdSCP113 singly perhaps is unable to alter the pathogen virulence. Another hypothesis is that the effectors initially function in virulence and subsequently are recognized by the host to induce immunity as a result of a long coevolutionary arms race, thereby also forcing a dual role during host-pathogen interactions. This may explain the finding that the double deletion mutant could reduce virulence, whereas the single gene deletion mutant did not reduce virulence, but can induce immunity.
In conclusion, through a high-throughput screen of all VdSCPs, our study demonstrated that transiently expressed VdSCP27, VdSCP113, and VdSCP126 in N. benthamiana participate in host-pathogen interactions. These three VdSCPs act as PAMPs to induce PTI responses differentially in a BAK1-dependent or BAK/SOBIR1-dependent manner, and cysteine residues are essential for the function of VdSCP126.
V. dahliae probably employs the effector VdCBM1 and other six VdSCP effectors to surmount the immunity triggered by these three VdSCPs.
Genes encoding VdSCP27, VdSCP113, and VdSCP126 are individually not essential for V. dahliae infection of host plants, but simultaneous deletion of both VdSCP27 and VdSCP126 contributes to reduced virulence on host plants. These results illustrate that the SCPs play a critical role through the intrinsic virulence function and suppress plant immunity functions. However, some of these have apparently been selected during the course of evolution of V. dahliae-host interactions and are thus recognized by the host to induce immunity.

| Growth of microbial and plant material
The V. dahliae wild-type strain Vd991 (highly virulent isolate from G. hirsutum) was cultured on potato dextrose agar (PDA) or in liquid Czapek medium for 7 days at 25 °C. The V. dahliae genetic transformants constructed in this study were cultured on PDA with 50 μg/ml hygromycin B. A B. cinerea wild-type strain was grown on PDA at 20 °C.
Agrobacterium tumefaciens GV3101 was cultured in Luria Bertani medium (tryptone 10 g, NaCl 10 g, yeast extract 5 g, water 1 L) at 28 °C for Both G. hirsutum and N. benthamiana were grown with a 14-hr light/10hr dark photoperiod in a greenhouse at the temperatures noted above.

| Bioinformatics analysis of VdSCPs
Sequences of the VdSCPs in this study were derived from the SCP predictions using the strain Vd991 reference genome (Chen et al., 2018). DNA sequences corresponding to each of the chromosomes of strain Vd991 were assembled based on alignment with the V. dahliae strain JR2 genome (de Jonge et al., 2013;Chen et al., 2018). Subcellular localization predictions of VdSCPs were performed using the Fungi model of WoLF PSORT (Horton et al., 2007) to identify the extracellular secretion characteristics, the signal peptides and signal peptide cleavage sites of VdSCPs were predicted using SignalP v. 4.1 (D-score cut-off set to 0.500) (Petersen et al., 2011), and the transmembrane domains were predicted by Phobius (Käll et al., 2007) and TMHMM 2.0 (Krogh et al., 2001). The conserved motifs of VdSCPs were analysed with the InterPro database (Apweiler et al., 2001).

| Plasmid construction and preparation
For transient expression in N. benthamiana, sequences with a FLAG tag fused to the C-terminus were cloned into potato virus X vector pGR107 (Jones et al., 1999) (Jacobs et al., 1997) to form the recombinant constructs (pSUC::SP VdSCP27 , pSUC::SP VdSCP113 , and pSUC::SP VdSCP126 ). To study the subcellular localization of VdSCP27, VdSCP113, and VdSCP126 in planta, the fulllength VdSCP27, VdSCP113 ,and VdSCP126 cDNAs were introduced into the 5′ end of the gene encoding GFP of the pBinGFP4 vector driven by a cauliflower mosaic virus 35S promoter to generate the recombinant expression vectors (pBin::VdSCP27-GFP, pBin::VdSCP113-GFP, and pBin::VdSCP126-GFP). For VIGS analyses, the genes BAK1 and SOBIR1 were amplified from N. benthamiana cDNA and integrated into the vector pTRV2 to construct pTRV2::BAK1 and pTRV2::SOBIR1, and the plasmid pTRV2::GFP was used as a control. For the targeted gene deletions in V. dahliae, the constructs (pGKO::VdSCP27, pGKO::VdSCP113, and pGKO::VdSCP126) were generated based on a previously described method (Liu et al., 2013). Briefly, approximately 1 kb of 5′ and 3′ flanking sequences of the three SCPs were amplified from V. dahliae genomic DNA and fused to a 1.8 kb hygromycinresistance cassette. The fusion product was introduced into the binary vector pGKO2-Gateway using a standard BP reaction (the cloned DNA fragment containing the attB site was recombined with the donor vector containing the attP site in vitro to produce the recombinant vector). To construct the double gene deletion mutant in V. dahliae, the sequences of VdSCP27, VdSCP113, and VdSCP126 accompanied by two flanking sequences were amplified from V. dahliae strain Vd991 genomic DNA and cloned into the binary vector pCOM (pCOM::VdSCP27, pCOM::VdSCP113, and pCOM::VdSCP126) that carries the geneticin resistance gene (Zhou et al., 2013), and were subsequently reintroduced to the ∆VdSCP113, ∆VdSCP126, and ∆VdSCP27 strains, respectively.

| Yeast signal sequence trap system
Yeast signal sequence trap system functional validation of the predicted signal peptide was performed as described previously (Jacobs et al., 1997). The plasmid constructs pSUC::SP VdSCP27 , pSUC::SP VdSCP113 , and pSUC::SP VdSCP126 were transformed into the yeast strain YTK12. Positive clones were confirmed by PCR using vector-specific primers (Table S6). The transformants were grown on CMD-W medium containing sucrose instead of glucose (0.67% yeast N base without amino acids, 0.075% tryptophan dropout supplement, 2% sucrose, 0.1% glucose and 2% agar). To assay invertase secretion, the positive transformants were incubated on YPRAA medium (1% yeast extract, 2% peptone, 2% raffinose, and 2 mg/ml antimycin A) containing raffinose as the carbohydrate source. YTK12 transformed with the functional signal peptide of Avr1b or an empty pSUC2 vector were used as positive and negative controls, respectively (Gu et al., 2011).

| VIGS in N. benthamiana
VIGS assays based on recombinant tobacco rattle virus (TRV) were performed as described previously (Liu et al., 2002). The plasmid constructs pTRV1, pTRV2::BAK1, pTRV2::SOBIR1, and pTRV2::GFP were introduced into A. tumefaciens GV3101. Agrobacterium strains harbouring either the pTRV2::BAK1 or pTRV2::SOBIR1 plasmid were combined with strains harbouring the pTRV1 vector in a 1:1 ratio and co-infiltrated into two primary leaves of 4-week-old N. benthamiana plants. The plasmid pTRV2::GFP was used as the control. The effectiveness of the VIGS assay was evaluated using the phytoene desaturase (PDS) gene as previously described (Liu et al., 2002). The silencing efficiencies of NbBAK1 and NbSOBIR1 were validated by quantitative reverse transcription PCR (RT-qPCR) analysis of RNA.
The experiment was performed three times with six plants for each TRV construct per experiment.

| ROS activity, cellulose deposition, and electrolyte leakage
ROS generation was detected using 3,3′-diaminobenzidine (DAB) solution (Solarbio) at 2 days after agroinfiltration of the corresponding gene in N. benthamiana leaves, as described previously (Bindschedler et al., 2006). Callose deposition was determined under fluorescence microscopy with a UV filter. The N. benthamiana leaves were harvested 2 days after agroinfiltration. The infiltrated leaf discs were destained with 75% ethanol at 37 °C and incubated in 150 mM phosphate buffer (pH 9.5) containing aniline blue (approximately 1% wt/vol) (Sigma-Aldrich) for 2 hr in the dark. The experiments were repeated three times and the relative amounts of ROS and callose deposition were calculated using the software ImageJ. Electrolyte leakage assays were performed as described previously (Oh et al., 2010). Ion conductivity was measured using a conductivity meter with Probe LE703 (Mettler-Toledo).

| Protection against disease development after inoculation with B. cinerea
A. tumefaciens was transformed with the expression cassettes for VdSCP27, VdSCP113, and VdSCP126, and the strains harbouring the individual expression cassettes were agroinfiltrated into the whole area of leaves of 4-week-old N. benthamiana (at a proper concentration to avoid induced cell death). At 12 hr post-infiltration, 5 μl of 2 × 10 6 conidia/ml of B. cinerea was placed on the infiltrated area.
The inoculated plants were placed in an incubator at 25 °C and 80% relative humidity. The diameters of the lesions were measured 48 hr after inoculation. Infection assays were performed using three leaves per plant and experiments were repeated three times.

| Gene expression analysis
To determine the expression of VdSCP27, VdSCP113, and VdSCP126 during infection, 2-week-old cotton and 4-week-old N. benthamiana seedlings were root-dip-inoculated with 5 × 10 6 conidia/ml V. dahliae, and the roots were harvested at different time points. The AxyPreP Multisource Total RNA Miniprep Kit (Axygen) was used to isolate total RNA. RT-qPCR was performed under the following conditions: an initial 95 °C denaturation step for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The N. benthamiana and V. dahliae EF-1α were used as endogenous references. Relative transcript levels among various samples were determined using the 2 −∆∆Ct method from three independent experiments (Livak and Schmittgen, 2001).
Unpaired Student's t tests were performed to determine statistical significance. The transient expression samples were also used for the expression of defence response genes induced by VdSCP27, VdSCP113, and VdSCP126 or suppressed by VdCBM1 with the same methods. Primers used for expression profiling are listed in Table S6.

| Immunoblotting
To verify protein production during transient expression in N. benthamiana, total proteins were extracted using the P-PER Plant Protein Extraction Kit (Thermo Scientific) and Protease Inhibitor Cocktail Kit (Thermo Scientific) from the agroinfiltrated N. benthamiana leaves 60 hr after inoculation, following the manufacturer's instructions. The proteins were separated using 15% sodium dodecyl sulphate polyacrylamide electrophoresis gels, and transient protein expression in N. benthamiana was assessed using anti-FLAG M2 antibody (Sigma) and detected using the Pierce ECL western blotting substrate (Thermo Scientific).
A. tumefaciens-mediated transformation of V. dahliae for gene deletion by homologous recombination was conducted as described previously (Mullins and Kang, 2001). The transformants were grown on PDA supplemented with 50 μg/ml hygromycin or 20 μg/ml geneticin for selection of targeted gene deletion mutant strains. Gene replacements in the targeted deletion mutants were verified by PCR using the appropriate primer sets (Table S6). Pathogenicity assays were performed on cotton and N. benthamiana seedlings as previously described (Gui et al., 2017) using a root-dipping method. For fungal biomass quantification, the roots of three plants were harvested at 3 weeks after inoculation and ground to a powder for genomic DNA extraction. The fungal biomass was quantified by quantitative PCR as described by Santhanam et al. (2013). Verticillium elongation factor 1-α (EF-1α) was used to quantify fungal colonization, and the cotton 18S rRNA gene and N. benthamiana EF-1α served as an endogenous plant control.
Unpaired Student's t tests were performed to determine statistical significance.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.